boost converter using ir2117 dc-dc

[jurc]

buck converter mosfet

Hi guys! I'm really desperate here; no matter what I do, my buck converter just keeps burning my FETs. Here's the story;

I have 230Vac input, then a bridge and a capacitor to get DC. Then I have a MOS-N-FET SPP20N60C3 which I drive with a gate driver via a transformer. I use PWM at 200kHz and 25% duty cycle. If I put a 100W bulb as a load, the circuit works perfectly, no problems whatsoever (circuit 1).

However, if I put a buck converter, the MOSFET gets destroyed (Rds = 0 and Rgs = 0) the second I turn on the PWM. To ilustrate, I plugin the 220V; the blub stays off. Then I start PWM and the thing goes to full power. (circuit 2)

I've tried using other MOSFETs (Infineon, Fairchild and ST), but the same thing happens. So I'm asking for ideas, why does this happen?

Jure

Added after 16 minutes:

BTW, this is the gate drive scheme that I'm using:

If I power the circuit with 30V, everything works perfectly; I get nice clean signals. At 320V however

Jure

 

[FvM]

problems in a buck converter

In think, that the source of problems can be clearly determined without measurements in operation, e.g. by increasing the input voltage through a variable transformer or a HV DC lab supply.

I see three possible reasons for failure:
- overcurrents during startup due to unsuitable pwm control
- overvoltages in pwm operation cause by parasitic circuit inductances
- the gate transformer's leak inductance is too high to achieve proper gate waveforms with the higher gate charge present in high voltage operation.

Each of these effects could cause an operation beyond the MOSFETs SOA (safe operating area).

Slowly increasing the voltage and current levels hopefully can reveal a problem before MOSFET permanent damage occurs.

 

[jurc]

buck diode recovery failure

Thank you for a good answer! I can rule out the overcurrents in startup, since I've tried slowly rising the PWM to 25% (from 0 to 1000ns in 1ns increments). Also tried with soft start (220ohm resistor + relay); same efect.

That parasitic and leakage inductance problem sounds good, but I don't know if there's anything I can do about it. The transformer is wound on and EPCOS N30 toroidal core and 10 turns of bifilar winding. I can't create a transformer with a lower leakage inductance.

What bothers me is, that the zenner diode does not protect the gate :?

Jure

 

[qolpa]

buck converter zener diode

hi jurc,

put a series lamp(2 or 3 series 50W lamp will work) to your mosfet and thus you will not damage your mosfet even your pwm signal reaches %100. then check your feedback circuit(or pwm circuit) and find what makes pwm %100.

regards

 

[FvM]

ir2117 applÄ°catÄ°on

The FET won't get damaged without a serious problem in circuit operation. I only made some guesses about possible reasons. There may be other problems, too.

It seems like you could already rule out the first and third choose. (If you don't expect overcurrents, that could happen later after succesful soft start). The said bifilar trigger transformer should be expected O.K., I think.

But anyway, you should look out for suspicious waveforms before fatal MOSFET damage occurs. This rqeuires suitable equipment, e.g. a differential probe to monitot the gate-source and drain-source waveforms. I don't see an indication for failing gate protection, by the way. The gate short can be caused by a massive output overcurrent as well.

 

[jurc]

buck converter does not start problem

Thanks! Yes, I'm currently lacking some equipment for the job. I'll be getting a variable transformer and an isolation transformer for the oscilloscope soon from a friend. Then, hopefully, the real debugging will start.

Thank you again!

 

[Tahmid]

gate drive transformer toroid leak inductance

Hi,
Zener diode did not protect your mosfet, as your problem not related to gate to source problem. You have enough protection in that aspect with the application of zener and with the application of 1k resister from gate to source, you have neutralize the miller effect also. Your circuit is perfect. But---- the problem is with your gate drive circuit – I think.
Your pulse Transformer is providing distorted/inadequate pulse to the gate of the mosfet and hence the mosfet is destroyed time and again. So, please do the following,
Instead of pulse transformer, use mosfet driver IR2117 or inverted IR2118. I have used those and are reliable. These are having Schmidt trigger inside and also bootstrap circuit for high side drive,which will make the input signal perfect for mosfet gate and I hope you will be relieved of your problem.
You can test the type of signal from the pulse transformer to the mosfet gate with Oscilloscope for being sure before discarding Pulse Transformer for gate drive.
In S.M.P.S. circuits, use your Oscilloscope intensively.

 

[jurc]

buck converter fet drive

Guys, I really need some more help here, as It'll won't be soon before I get the equiptment.
I made an inverted buck converter; the FET is on the low side, so there is NO gate drive transformer or anything; just plain gate driver IC and an 2R2 resistor. Same results. The FET just gets destroyed as soon as PWM is turned on.

So this gets me thinking that the gate isn't the problem, maybe the Vds is. If the diode is too slow, then when the FET turns of, it induces very high voltage. Though the diode is 15ETX06 which is rated 35ns Trr; they don't get much faster than this.

Other thing I've tried is parasitic inductance of the wires. I've added in simulation two 100nH inductors:

And this does produce some high spikes on Vds:

So what can I do? Different diode? Should I try with a snubber or a TVS diode across the FET to protect it?

Thank you guys for all your help! I'm really strugling here with this

 

[Tahmid]

buck converter high side driver circuit

Hi jurc,
One more thing for your checking. I have studied the datasheet of the mosfet you have used in the first circuit and find that the reverse recovery period of the body diode is 500-800 ns. Considering your high freq. that is 200khz, it seems to be much slower and hence mosfet may be destroyed for cross conduction. So, you can discard the use of the body diode of the mosfet and can use an external diode with reverse recovery period < 70 ns. Thanks.

 

[FvM]

buck converter saturating choke

The reverse recovery behaviour of the body diode is meaningless, when the diode isn't forward biased in a circuit. Simply forget about this point.

It's basically a good idea to remove the trigger transformer by changing the topology.

100 nH parasitic inductances are unsuitable, if a similar inductance is present in a 200 kHz switcher, you have to redesign the geometry, I think. I guess, you simply don't know the real circuit inductance.

Regarding FET overvoltage behaviour, please consider, that your device has already an internal protection. So you can't damage it by some overvoltage, you have to supply sufficient power to exceed the protection diode rating, if FET overvoltage would be the problem.

Did you verify, that the diode is still working, particularly is standing full reverse voltage after the FET has been damaged? The diode is much more exposed to a possible overvoltage damage, because it has no internal protection (a diode can't have it). So one possible damage scenario is breakdown of the diode and following pulse power overload of the FET. Although I would expect the diode to be damaged too, it's possibly alive at first sight.

If diode breakdown would trigger the damage process (still a guess), then a snubber at the diode could help. Reducing parasitic inductance of the commutating path can eliminate the problem.

In any case, if you don't have the tools to measure waveforms in the critical voltage/current range, it's rather difficult to clarify the issue.

 

[jurc]

ir2118, scheme

Thank you! I'll test tomorrow if the diode is still OK. I've indeed been testing with the same diode.

I've just tired another circuit. Consider the above buck converter, but remove the inductor and output capacitor. Works perfectly. Add those two and the FET gets destroyed. So the inductor is doing something.

Jure

 

[ahoss2000]

zener buck

I think the problem is with inductor. Tell us what are you using: Core size/material and number of turns. There is a large voltage across the inductor, and it needs to have a good size core/torroid with many turns.

Let us know.

 

[jurc]

buck converter fet

Thank you for the idea. To go a bit back, I've tested the diode, and It's fine; withstands full reverse voltage. Forward conduction is also OK.

As for the inductor; I'm using the ETD39 core made of Epcos N97 material.
https://uk.farnell.com/epcos/b66363gx197/ferrite-core-half-etd39-n97/dp/1422748

It has 12 turns of 2x1mm (diameter) copper wire. I made a few simple calculations and at the current the 100W bulb draws, it's well below saturation. Here's a shot:

I also tried with only one 1mm wrire and the effect is the same. Anyways, what could be wrong with the construction of the inductor, that it could cause the FET to get destroyed?

Thank you again all!

 

[FvM]

buck converter variable mosfet

I made a few simple calculations and at the current the 100W bulb draws, it's well below saturation.

I hope so. What's the nominal current? It's very unusual however to use an ungapped core for a DC choke.

 

[Tahmid]

diode reverse recovery buck converter

If slight airgap is provided to the inductor, circuit will be stabilized as freq. is 200khz. Without airgap, ETD 39core may not act according to your requirement as this core is not made for Inductor purpose. I think it will be worth trying for you. Thanks.

 

[jurc]

Re: FET in buck converter

Hi guys,
I'm still having the same problems. Here's a screenshot from the oscilloscope of Vds of the FET:

This is when converter operates at 12V. I can only imagine what happens at 320V. The snubber does reduce the ringing considerably, however I have to use 10nF and 2.2R resistor (1W), which starts to overheat at about 25V. I don't know how to make a snubber for 320V that's reasonably small. Any ideas?

Thanks for all your help guys.

Regards!

 

[analog_match]

FET in buck converter

I am wondering if reverse recovery of 15ETX06 is causing FET to operate out of SOA?

 

[jurc]

Re: FET in buck converter

Well, It's a good diode, 22ns Trr; they don't get much faster. Hm... I'll put it like this; can anyone tell me what would a good diode be for my application? Name one and I'll get it from farnell or digikey to test it. I'm really getting low on options right now

regards

 

[analog_match]

FET in buck converter

jurc,
the forum would benefit from knowing what you did to ultimately fix this issue?
Thanks

 

[marthinus]

Re: FET in buck converter

Your inductor is saturating.It causes massive current through the fet.